We conducted an online choice survey to inform advice to the Tasman District Council on setting management objectives for multiple uses and values across several catchments. One sub-sample was recruited via a survey company and a second via a public call with prize draw. From a survey with unlabelled choice sets for three rivers, we estimated a separate model for each river. Coefficients for natural character, fish & fishing, local jobs and cost were generally significant and had expected signs. Coefficients on swimming and boating attributes were weak even though 68% and 31%, respectively, of the panel sample reported engaging in these activities. Levels of attributes varied for the three rivers and, together with within-sample variation, made cross-river comparisons challenging. Latent class analysis was used to assess non-attendance with interesting results. The panel and public sample results highlighted the effect of random versus non-random sampling.